Monday, October 13, 2014

Year-end Analysis of Racing in the MidAtlantic Bicycle Racing Association


I have a much nicer PDF of this analysis--ask for it in the comments section. And it has all the notes

1 Introduction

This report attempts to summarize the attendance trends over the past nine years of racing in the Mid-Atlantic Bicycle Racing Association–MABRA. I hope it is a useful resource for both racers and promoters. In drawing conclusions from these tables and graphs, you should be aware that the data sets are small, and that hidden variables almost certainly influence the trends. However, I believe that some data is better than
no data at all. I am always looking for ways to improve this document, and I consider all suggestions.
As always, this report is an incremental modification to the report of previous years. To keep the main body of the report to a manageable and readable length, the supporting information appears as a collection of endnotes that follows the last section.( just ask for the PDF version if you want this)

2 What is MABRA?

MABRA, the Mid Atlantic Bicycle Racing Association, is an association of event-promoting clubs, based loosely around the Washington, DC-Baltimore, Md area. The map in Figure 1 shows the traditional geographic outline of MABRA, and the approximate locations of its member clubs in late 2014. The diameter of the circle is proportional to the number of USAC-licensed members in the club. (see note 1)  
Figure 1: Map of the traditional geographic boundaries of MABRA. Circle position indicates the location of a member club in 2014. Circle diameter is proportional to the number of members.

Most clubs have many members who are unlicensed. In addition, most clubs have many members, mostly Category 5, who are listed as “unattached” but who have actually joined the club. Figure 2 shows a zip-code-by-zip-code map of MABRA in which the size and color of the plotted point indicate the number of licenses in that zip code. The ten most populated zip codes in MABRA are all in DC or the counties that border DC. For the US as a whole, 7 of the top 10 zip codes are Boulder or its surrounding towns. 

Figure 2 shows a zip-code-by-zip-code map of MABRA in which the size and
color of the plotted point indicate the number of licenses in that zip code. 
Figure 2: Zip-code map of MABRA racer residences. Color is proportional to the number of licensed racers in the zip code. Size of spot increases with increasing numbers. Zip codes with no licensed racers are show as gray dots.
Add caption

3 Demographics

Figure 3 plots the demographic data by age broken out by gender and racing category
for all MABRA riders in 2014. Figure 4 slices puts the data of Figure 3 into separate
panels by category.
Figure 3 Distribution of age of MABRA racers by gender and category.
Figure 4 Distribution of age of MABRA races by category and gender.

Figure 5 compares the age distribution of all USCF-licensed riders in 1989, (note 2) and
2005 to 2014. 3 In the 25 years since 1989, the median rider age has increased 14 years,
from 27 to 41 (computed for men, though Figure 5 shows both genders). Note also the asymmetry of the 1989 distribution towards a larger fraction of young riders is much larger than the corresponding asymmetry in 2005. Note also that from 2005 to 2014 the distribution broadens both to younger and
to older ages. In 2014, a significant hump appears for racers less than 20 years old. 

Historical USACycling membership demographics

Figure 6 shows that the median age of licensed racers, except for category 5 has steadily increased since 2005. Interestingly, the median Category 5 age is basically constant. MABRA is not adding younger racers. Note also that the median age for women is basically unchanging. 

Figure 6. Historical median racing age by category and gender. 

4 Historical USAC and MABRA data

Figure 7 plots historical data on USACycling membership through the years, as well as some formative events in American cycling history. (note 4) USCF membership skyrocketed during the 1980s before reaching a plateau in 1990 and then declining. It was not until 2005 that membership exceeded the 1992 peak. What caused the plateau? Why is membership increasing again after 2003? It’s hard not to argue that it’s the “Lance Armstrong effect.” But then, why does the membership continue to grow even after all the doping revelations of the past several years?

Figure 7. Historical USACycling membership data. 

 Figure 8 plots the analogous data for total MABRA memberships. After three years of stagnant growth, membership increased again in 2014.
Figure 8 Historical MABRA membership data.

Race Attendance–Who is actually racing?

Figure 9 shows the fraction of of the total number of races entered by everyone who competed in a MABRA or Virginia race I was able to locate.5 This data set includes every racer with an address in MABRA, and every racer, regardless of address, who appears in a result from 2013. Note that licensed racers who competed in zero, one, or two races comprise more than half of the total.
Figure 9 Histogram of the fraction (as percent) of races in MABRA/VA entered by each licensed racer

Figure 10 is a similar histogram but that only considers MABRA racers (i.e. licensed racers whose address is in the geographic boundaries of MABRA). Here, more than one quarter of all racers do not appear in any results. And half of the licenses racers competed in two or fewer events in 2013.

Histogram of the fraction (as percent) of of races in MABRA entered by each licensed MABRA racer.

Figure 11 takes that data and plots it as a cumulative distribution beginning with the racer, regardless of address, with the most races in 2013.6 The most interesting conclusion is that 10 % of licensed racers account for 40 % of the attendance. This result confirms the anecdotal evidence that you see the same people at races every week.

Figure 11. Cumulative distribution of race attendance. (Results do not reach unity because unknown licenses are not considered.)


forgot it--ask for the PDF version.